The hardest thing on a bad-signal job is not fixing it. It is deciding you are not going to.
I have stood on a ridge for six minutes trying to get a picture to hold while a homeowner sat in a break room watching a spinner. That is worse than any version of calling it off. It burned her lunch break, it made me look like I did not know my own tools, and I still had to send her the footage afterward anyway.
Knowing when to reschedule an inspection, or more often when to change the format rather than the date, is a skill. Here is how I make the call. It starts with deciding before the ladder, not on the roof.
Almost every bad outcome I have had came from committing before I knew. The homeowner is already on the link, expectations are set, and now every decision is a retreat.
So the check happens on the ground. Wi-Fi off, upload test twice with fifteen seconds between, look at the real signal numbers instead of the bars, and ask the homeowner where they will be watching from. I go through that pre-flight in detail elsewhere. The point here is that it produces a decision, and the decision gets made while nothing is riding on it.
If sustained upload is under roughly 800 kbps, or the two tests disagree wildly, or there is visible packet loss, I do not go live. That is not a maybe.
The three things that mean stop
The link will not hold. Not slow, unstable. Slow I can work with by moving deliberately and holding on details. Unstable I cannot, because it does not degrade gracefully, it freezes. A homeowner watching a frozen frame of a pipe boot for forty seconds is not getting documented, they are getting frustrated.
The phone is already in trouble. If the device is hot and dimming before I have climbed, on a black roof in August, I know how the next twenty minutes go. A phone with no thermal headroom at the start of a walk will not finish one, and clipping a battery pack to it does not help, because most phones stop accepting a charge once they are that hot.
The homeowner cannot actually watch. This is the one that gets skipped. Your uplink is irrelevant if they are on a train, in a parking garage, on a hotel network, or driving. Ask. If they say they are between meetings and might have to step out, that is a scheduling problem, not a network problem, and no amount of signal fixes it.
Reschedule less often than you think
Here is the part that matters commercially. Most of the time, when to reschedule an inspection is the wrong question. The right question is what format the job runs in.
You are already there. The ladder is on the truck. The homeowner arranged their day. Driving away and rebooking costs you a second trip and costs them another window, and you are gambling that the signal at that address will be different on a Thursday than it was on a Tuesday. Frequently it will not be, because the cause is the tower, not the day.
So the default alternative is not cancel. It is record and send.
Do the same walk you would have done live. Same order, same narration, same holds on the details. Record it locally at full capture quality, which is not limited by uplink at all, and let it upload when you are back in coverage. Then call the homeowner from the truck, or meet them at the door, and go through it with the footage playing on your phone or theirs. They still get the walk, they still get their questions answered, and they get a cleaner picture than a struggling live stream would have given them.
What you lose is the ability for them to ask a question while you are still standing on the condition. That is a real loss and I do not pretend otherwise. It is not worth trading twenty minutes of frozen video for.
Say it before it happens
The whole thing turns on when you tell them.
Before: "Service is thin out here, so instead of streaming I am going to record the whole walk and go through it with you the second I am down. You will actually see more this way." That is a professional making a call.
After: "Sorry, the video kept freezing." That is an apology, and it lands as an excuse whether or not it is one.
Same facts, and homeowners react to them completely differently. The only variable is whether you knew first. Which is the entire argument for spending sixty seconds in the driveway.
I also tell them what the middle case looks like, because most marginal jobs land there. Between roughly 1 and 3 Mbps up I will go live, and I will say up front that the picture is going to be soft, that I am going to move slowly and hold on things, and that they will get a clean recording afterward regardless. Nobody is upset by a soft picture they were warned about. Everybody is upset by one they were not.
The safety version of this decision
Worth saying plainly because it is the actual risk on this topic. Do not troubleshoot a phone on a slope.
Weak signal makes people wander. You start walking the ridge holding the phone up, drifting toward edges, looking at a screen instead of your feet, one hand occupied. Every one of those is a bad idea individually and you are doing four at once, on a surface that may be hot, dusty, or wet.
If the stream is not working, get down and fix it, or get down and change the plan. Nothing on a residential roof is worth a fall, and no homeowner has ever asked me to take a risk for a live picture. They just want to know what is on their roof and to have a reason to believe it.
The rule I actually use
If I can hold a stable stream, I go live. If I cannot, I record the same walk and review it with them within the hour. If the phone or the weather or the light is against me, or the homeowner cannot watch properly, I reschedule the whole thing and say so before anyone has been promised a window.
The underlying mechanics of why some roofs are hopeless and some are just slow are in the piece on streaming at one bar, and they are worth understanding, because they turn this from a guess into a decision.
The best time to learn where your own line is, is on a roof where nobody is waiting. Run a test at the edge of your worst service area, push it until it breaks, and see what the failure actually looks like from both ends of the session.
InspectStream records what an inspector observed. It is not a public adjusting service and does not prepare, negotiate, or advise on insurance claims.